AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

Learning Fair Policies in Decentralized Cooperative Multi-Agent Reinforcement Learning

Matthieu Zimmer, Claire Glanois, Umer Siddique, Paul Weng

ICML 2021 · Multi-agent systems and games › Cooperative multi-agent RL · approach: Decentralized actor-critic MARL · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use the SOTO two-subnetwork architecture instead of BasicPPO with GGF or α-fairness · Matthew EffectPareto efficiency and equityImproved
Use SOTO with GGF instead of FENCooperative MARL · Matthew Effect, CLDEtotal income and minimum resources collectedImproved
Use SOTO with α-fairness instead of FENCooperative MARL · Matthew Effectminimum resources collectedImproved
Use SOTO with α-fairness instead of FENCooperative MARL · Matthew Effectcoefficient of variationMixed
Use a centralized critic with GGF instead of SOTO with GGFCooperative MARL · Job Schedulingcoefficient of variationImproved
Use WQMIX instead of other compared methodsCooperative MARL · Job Schedulingcoefficient of variationHurt
Use cooperation instead of independent selfish policiesTraffic-light control · SUMOglobal waiting timeImproved
Use SOTO with GGF instead of other compared methodsTraffic-light control · SUMOglobal waiting timeImproved
Use fully decentralized SOTO with GGF instead of FEN or WQMIXTraffic-light control · SUMOcoefficient of variationImproved
Use SOTO with GGF instead of FEN or a random policyData-center control · Continuous actionscoefficient of variationImproved
Remove the trained self-oriented policy by using a randomly initialized one (ablation)SOTO · Matthew Effectsocial welfare optimizationHurt
Replace incremental switching with pretraining the self-oriented policy first (ablation)SOTO · Matthew Effect, CLDEdata efficiencyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy